public static List<string> GetListFromLexRecord(LexRecord lexRecord, int contentType)
{
List<string> outList = new List<string>();
switch (contentType)
{
case 2:
outList = lexRecord.GetSpellingVars();
break;
case 5:
outList = lexRecord.GetVariants();
break;
case 6:
outList = lexRecord.GetNominalizations();
break;
case 7:
outList = lexRecord.GetAbbreviations();
break;
case 8:
outList = lexRecord.GetAcronyms();
break;
}
return outList;
}
/**
* extract information about acronyms from NIH record, and add to a
* simplenlg WordElement.
*
* <P>
* Acronyms are represented as lists of word elements. Any acronym will have
* a list of full form word elements, retrievable via
* {@link LexicalFeature#ACRONYM_OF}
*
* @param wordElement
* @param record
*/
private void addAcronymInfo(WordElement wordElement, LexRecord record)
{
// NB: the acronyms are actually the full forms of which the word is an
// acronym
List <string> acronyms = record.GetAcronyms();
if (!acronyms.isEmpty())
{
// the list of full forms of which this word is an acronym
List <INLGElement> acronymOf = wordElement
.getFeatureAsElementList(LexicalFeature.ACRONYM_OF);
// keep all acronym full forms and set them up as wordElements
foreach (var fullForm in acronyms)
{
if (fullForm.contains("|"))
{
// get the acronym id
string acronymID = fullForm.substring(
fullForm.indexOf("|") + 1, fullForm.length());
// create the full form element
WordElement fullFormWE = this.getWordByID(acronymID);
if (fullForm != null)
{
// add as full form of this acronym
acronymOf.add(fullFormWE);
// List<NLGElement> fullFormAcronyms = fullFormWE
// .getFeatureAsElementList(LexicalFeature.ACRONYMS);
// fullFormAcronyms.add(wordElement);
// fullFormWE.setFeature(LexicalFeature.ACRONYMS,
// fullFormAcronyms);
}
}
}
// set all the full forms for this acronym
wordElement.setFeature(LexicalFeature.ACRONYM_OF, acronymOf);
}
// if (!acronyms.isEmpty()) {
//
// string acronym = acronyms.get(0);
// // remove anything after a |, this will be an NIH ID
// if (acronym.contains("|"))
// acronym = acronym.substring(0, acronym.indexOf("|"));
// wordElement.setFeature(LexicalFeature.ACRONYM_OF, acronym);
// }
return;
}
public static void SetItemInListInLexRecordAt(LexRecord lexRecord, int contentType, string item, int index)
{
switch (contentType)
{
case 2:
lexRecord.GetSpellingVars()[index] = item;
break;
case 5:
lexRecord.GetVariants()[index] = item;
break;
case 6:
lexRecord.GetNominalizations()[index] = item;
break;
case 7:
lexRecord.GetAbbreviations()[index] = item;
break;
case 8:
lexRecord.GetAcronyms()[index] = item;
break;
}
}
private static void GetLexRecord(Node node, LexRecord lexRecord, CatEntry catEntry)
{
if (node.NodeType == XmlNodeType.Element)
{
if (node.Name.Equals("base"))
{
lexRecord.SetBase(getValue(node));
}
else if (node.Name.Equals("eui"))
{
lexRecord.SetEui(getValue(node));
}
else if (node.Name.Equals("cat"))
{
lexRecord.SetCat(getValue(node));
}
else if (node.Name.Equals("spellingVars"))
{
lexRecord.SetSpellingVar(getValue(node));
}
else if (node.Name.Equals("acronyms"))
{
lexRecord.GetAcronyms().Add(getValue(node));
}
else if (node.Name.Equals("abbreviations"))
{
lexRecord.GetAbbreviations().Add(getValue(node));
}
else if (node.Name.Equals("annotations"))
{
lexRecord.GetAnnotations().Add(getValue(node));
}
else if (node.Name.Equals("signature"))
{
lexRecord.SetSignature(getValue(node));
}
else if (node.Name.Equals("verbEntry"))
{
catEntry = getVerbValues(node);
lexRecord.SetCatEntry(catEntry);
}
else if (node.Name.Equals("auxEntry"))
{
catEntry = getAuxValues(node);
lexRecord.SetCatEntry(catEntry);
}
else if (node.Name.Equals("modalEntry"))
{
catEntry = getModalValues(node);
lexRecord.SetCatEntry(catEntry);
}
else if (node.Name.Equals("nounEntry"))
{
catEntry = getNounValues(node);
lexRecord.SetCatEntry(catEntry);
}
else if (node.Name.Equals("pronEntry"))
{
catEntry = getPronValues(node);
lexRecord.SetCatEntry(catEntry);
}
else if (node.Name.Equals("adjEntry"))
{
catEntry = getAdjValues(node);
lexRecord.SetCatEntry(catEntry);
}
else if (node.Name.Equals("advEntry"))
{
catEntry = getAdvValues(node);
lexRecord.SetCatEntry(catEntry);
}
else if (node.Name.Equals("detEntry"))
{
catEntry = getDetValues(node);
lexRecord.SetCatEntry(catEntry);
}
}
for (Node child = node.FirstChild; child != null; child = child.NextSibling)
{
GetLexRecord(child, lexRecord, catEntry);
}
}
public static bool Check(LexRecord lexRecord, HashSet <string> notBaseFormSet)
{
bool validFlag = true;
List <string> acrList = lexRecord.GetAcronyms();
string acrCat = lexRecord.GetCategory();
for (int i = 0; i < acrList.Count; i++)
{
string acr = (string)acrList[i];
int index1 = acr.IndexOf("|", StringComparison.Ordinal);
string acrCit = "";
string acrEui = "";
if (index1 > 0)
{
acrCit = acr.Substring(0, index1);
acrEui = acr.Substring(index1 + 1);
}
else
{
acrCit = acr;
}
string citCat = acrCit + "|" + acrCat;
HashSet <string> euisByCit = CrossCheckDupLexRecords.GetEuisByCitCat(citCat);
if (euisByCit == null)
{
if (acrEui.Length > 0)
{
acrList[i] = acrCit;
validFlag = false;
ErrMsgUtilLexicon.AddContentErrMsg(5, 3, acr + " - None", lexRecord);
}
else
{
validFlag = false;
if (!notBaseFormSet.Contains(citCat))
{
ErrMsgUtilLexicon.AddContentErrMsg(5, 4, acr + " - New", lexRecord);
}
}
}
else if (euisByCit.Count == 1)
{
List <string> euiList = new List <string>(euisByCit);
string newEui = (string)euiList[0];
if (acrEui.Length > 0)
{
if (euisByCit.Contains(acrEui) != true)
{
validFlag = false;
ErrMsgUtilLexicon.AddContentErrMsg(5, 6, acr + " - " + newEui, lexRecord);
}
}
else
{
string newAcr = acr + "|" + newEui;
acrList[i] = newAcr;
validFlag = false;
ErrMsgUtilLexicon.AddContentErrMsg(5, 5, acr + " - " + newEui, lexRecord);
}
}
else
{
List <string> euiList = new List <string>(euisByCit);
if (acrEui.Length > 0)
{
if (euisByCit.Contains(acrEui) != true)
{
validFlag = false;
ErrMsgUtilLexicon.AddContentErrMsg(5, 8, acr + " - " + euiList, lexRecord);
}
}
else
{
validFlag = false;
ErrMsgUtilLexicon.AddContentErrMsg(5, 7, acr + " - " + euiList, lexRecord);
}
}
}
return(validFlag);
}